Recent content by DeepMarket

I was just stating that in the threaded applications that I have written in Java and C++, I did not have to tell the OS which CPU to use. Be it a dual processor or a 64 processor machine - the OS always took care of it. Now, that was with Solaris, Dynix, Linux - I don't know how it is with...

THE most important thing is that you are happy :) Of course, I get excited over a new can of air to clean my computers...

Can you help me out and point to which programming languages you can do this? I cannot do this in Java. Years and years ago when I wrote C++ for Solaris - I couldn't do it with C++. Granted, I have not done any threaded applications for Windows, but I thought all the that was taken care of...

I didn't realize that - I was always under the impression that the OS took care of assigning the thread to a processor and that it would happen in a non-deterministic fashion depending on system resources. Good to know.

You bring up a key point. This is the "magic" behind any and all successfully implemented learning algorithm - getting the data into a format that makes it easy for the technique to tease out meaningful patterns. I remember when I first threw raw stock data at a neural network back in the...

If you can get a full T-1 for less than $500 - jump on it. That is a bargin. No consumer broadband (cable/dsl) can compare in terms of speed and reliability (ah - depending on provider...).

If you are familiar with Digg, del.icio.us, or Furl - then you already understand the social bookmarking concept. If not, these sites help users to bookmark sites that interest them and share those bookmarks with other. As with anything, the success in one area can lead to others...

I wouldn't recommend automatically trading such pattern matches - perhaps you are better off being alerted to a "potential pattern". The reason I say this is, I have been looking at automated ways to identify patterns in stock charts. Here you can see automated identification of "Head and...

June 6, 2006

Also, depending on how many data streams you are looking at, you may want to consider coarse grained parallelism. For instance, if you are looking at multiple streams of data, you might be better off building a cluster of super-cheap AMD machines and calculating each data stream on its own...

I will second that. SpinRite is awesome and has been around since the 80's (obviously it has been upgraded as HD technology has evolved). It can take a looooooooong time to get things corrected, but if it is correctable - SpinRite is your best option for less than $100.

Sorry - didn't mean to sound like I was looking for a handout, I was more interested in your perspective rather than the actual answer. I agree 100% - figuring it out oneself is the best route - you can see some of my research over at deepmarket.com.

That is an interesting way to look at drawdowns - I had not given much thought to it that way. Have you come up with an alternative measure or are you pointing out the problem?

No - I don't argue with success. If you are - excellent! It is just my experience that I have not come across anything out there like you are describing - it is just too difficult. I would certainly love to hear about your methods and results.